Gate Dielectric Chemical Structure−Organic Field-Effect Transistor

Electron trapping due to silanol and carbonyl functionalities at the semiconductor−dielectric interface is identified as the principal origin of the...
1 downloads 4 Views 1MB Size
Published on Web 09/09/2006

Gate Dielectric Chemical Structure-Organic Field-Effect Transistor Performance Correlations for Electron, Hole, and Ambipolar Organic Semiconductors Myung-Han Yoon, Choongik Kim, Antonio Facchetti,* and Tobin J. Marks* Contribution from the Department of Chemistry and the Materials Research Center, Northwestern UniVersity 2145 Sheridan Road, EVanston, Illinois, 60208 Received May 11, 2006; E-mail: [email protected]; [email protected]

Abstract: This study describes a general approach for probing semiconductor-dielectric interfacial chemistry effects on organic field-effect transistor performance parameters using bilayer gate dielectrics. Organic semiconductors exhibiting p-/n-type or ambipolar majority charge transport are grown on six different bilayer dielectric structures consisting of various spin-coated polymers/HMDS on 300 nm SiO2/p+-Si, and are characterized by AFM, SEM, and WAXRD, followed by transistor electrical characterization. In the case of air-sensitive (generally high LUMO energy) n-type semiconductors, dielectric surface modifications induce large variations in the corresponding OTFT performance parameters although the film morphologies and microstructures remain similar. In marked contrast, the device performance of air-stable n-type and p-type semiconductors is not significantly affected by the same dielectric surface modifications. Among the bilayer dielectric structures examined, nonpolar polystyrene coatings on SiO2 having minimal gate leakage and surface roughness significantly enhance the mobilities of overlying air-sensitive n-type semiconductors to as high as ∼ 2 cm2/(V s) for R,ω-diperfluorohexylcarbonylquaterthiophene polystyrene/SiO2. Electron trapping due to silanol and carbonyl functionalities at the semiconductor-dielectric interface is identified as the principal origin of the mobility sensitivity to the various surface chemistries in the case of n-type semiconductors having high LUMO energies. Thiophene-based n-type semiconductors exhibiting similar film morphologies and microstructures on various bilayer gate dielectrics therefore provide an incisive means to probe TFT performance parameters versus semiconductor-dielectric interface relationships.

Introduction

Field-effect-active organic semiconductors (OSCs)1,2 are of great interest for use in low-cost/disposable electronic products such as smart cards and radio frequency identification (RFID) (1) For recent reviews on this topic, see: (a) Chabinyc, M.; Loo, Y.-L. J. Macrom. Sci. Polym. ReV. 2006, 46, 1-5. (b) Dodabalapur, A. Nature 2005, 434, 151. (c) Sirringhaus, H. AdV. Mater 2005, 17, 2411. (d) Newman, C. R.; Frisbie, C. D.; da Silva Filhoqq, D. A.; Bredas, J.-L.; Ewbank, P. C.; Mann, K. R. Chem. Mater. 2004, 16, 4436. (e) Horowitz, G. J. Mater. Res. 2004, 19, 1946. (f) Dimitrakopoulos, C. D.; Malenfant, P. R. L. AdV. Mater. 2002, 14, 99. (g) Rogers, J. A.; Bao, Z.; Katz, H. E.; Dodabalapur, A. Thin-Film Transistors; Kagan, C., Andry R. P., Eds; Marcel Dekker, Inc.; New York, USA, 2003; pp 377-425. (2) (a) Kim, D. H.; Han, J. T.; Park, Y. D.; Jang, Y.; Cho, J. H.; Hwang, M.; Cho, K. AdV. Mater. 2006, 18, 719. (b) Tulevski, G. S.; Miao, Q.; Afzali, A.; Graham, T. O.; Kagan, C. R.; Nuckolls, C. J. Am. Chem. Soc. 2006, 128, 1788. (c) Ponomarenko, S. A.; Kirchmeyer, S.; Elschner, A.; Alpatova, N. M.; Halik, M.; Klauk, H.; Zschieschang, U.; Schmid, G. Chem. Mater. 2006, 18, 579. (d) Ando, S.; Nishida, J.; Tada, H.; Inoue, Y.; Tokito, S.; Yamashita, Y. J. Am. Chem. Soc. 2005, 127, 5336. (e) Ofuji, M.; Lovinger, A. J.; Kloc, C.; Siegrist, T.; Maliakal, A. J.; Katz, H. E. Chem. Mater. 2005, 17, 5748. (f) Wu, Y.; Ong, B. S. J. Am. Chem. Soc. 2005, 127, 3266. (g) Murphy, A. R.; Liu, J.; Luscombe, C.; Kavulak, D.; Frechet, J. M. J.; Kline, R. J.; McGehee, M. D. Chem. Mater. 2005, 17, 4892. (h) Locklin, J.; Li, D.; Mannsfeld, S. C. B.; Borkent, E.-J.; Meng, H. Advincula, R.; Bao, Z. Chem. Mater. 2005, 17, 3366. (i) Payne, M. M.; Parkin, S. R.; Anthony, J. E. J. Am. Chem. Soc. 2005, 127, 8028. (j) Swartz, C. R.; Parkin, S. R.; Bullock, J. E.; Anthony, J. E.; Mayer, A. C.; Malliaras, G. G. Org. Lett. 2005, 7, 3163. (k) Payne, M. M.; Parkin, S. R.; Anthony, J. E.; Kuo, C.-C.; Jackson, T. N. J. Am. Chem. Soc. 2005, 127, 4986. (l) Sakamoto, Y.; Suzuki, T.; Kobayashi, M.; Gao, Y.; Fukai, Y.; Inoue, Y.; Sato, F.; Tokito, S. J. Am. Chem. Soc. 2004, 126, 8138. (m) Kunugi, Y.; Takimiya, K.; Toyoshima, Y.; Yamashita, K.; Aso, Y.; Otsubo, T. J. Mater. Chem. 2004, 14, 1367. 10.1021/ja063290d CCC: $33.50 © 2006 American Chemical Society

tags,3 as well as in flexible display driver circuits,4 nonvolatile memories,5 and sensors.6 Indeed, amorphous and polycrystalline films of several OSCs exhibit hole or electron carrier mobilities (3) (a) Yoo, B.; Jung, T. O.; Basu, D.; Dodabalapur, A.; Jones, B. A.; Facchetti, A.; Wasielewski, M. R.; Marks, T. J. Appl. Phys. Lett. 2006, 88, 082104. (b) Subramanian, V.; Chang, P. C.; Lee, J. B.; Molesa, S. E.; Volkman, S. K. IEEE Trans. Comp. Pack. Tech. 2005, 28, 742. (c) Baude, P. F.; Ender, D. A.; Haase, M. A.; Kelley, T. W.; Muyres, D. V.; Theiss, S. D. Appl. Phys. Lett. 2003, 82, 3964. (d) Drury, C. J.; Mutsaers, C. M. J.; Hart, C. M.; Matters, M.; de Leeuw, D. M. Appl. Phys. Lett. 1998, 73, 108. (e) Gamota, D. R.; Brazis, P.; Kalyanasundaram, X.; Zhang, J. Printed Organic and Molecular Electronics; Kluwer Academic Publishers: New York, USA, 2004. (4) (a) Street, R. A.; Wong, W. S.; Ready, S. E.; Chabinyc, M. L.; Arias, A. C.; Limb, S.; Salleo, A.; Lujan, R. Mater. Today (Oxford, U. K.) 2006, 9, 32. (b) Zhou, L.; Wanga, A.; Wu, S.-C.; Sun, J.; Park, S.; Jackson, T. N. Appl. Phys. Lett. 2006, 88, 083502/1. (c) Zyung, T.; Kim, S. H.; Lim, S. C.; Lee, J. H.; Chu, H. Y.; Lee, J.-I.; Oh, J. Y. J. Korean Phys. Soc. 2006, 48, S111. (d) Oh, S. Y.; Lee, J.-Y Mol. Cryst. Liq. Cryst. 2006, 444, 211. (e) Chabinyc, M. L.; Salleo, A. Chem. Mater. 2004, 16, 4509. (f) Sheraw, C. D.; Zhou, L.; Huang, J. R.; Gundlach, D. J.; Jackson, T. N.; Kane, M. G.; Hill, I. G.; Hammond, M. S.; Campi, J.; Greening, B. K.; Francl, West, J. J. Appl. Phys. Lett. 2002, 80, 1088. (g) Mach, P.; Rodriguez, S. J.; Nortrup, R.; Wiltzius, P.; Rogers, J. A. Appl. Phys. Lett. 2001, 78, 3592. (h) Klauk, H.; Gundlach, D. J.; Nichols, J. A.; Jackson, T. N. IEEE Trans. Electron DeVices 1999, 46, 1258. (5) Naber, R. C. G.; de Boer, B.; Blom, P. W. M.; de Leeuw, D. M. Appl. Phys. Lett. 2005, 87, 203509/1. (6) (a) Someya, T.; Sekitani, T.; Iba, S.; Kato, Y.; Sakurai, T.; Kawaguchi, H. Mol. Cryst. Liq. Cryst. 2006, 444, 13. (b) Wang, L.; Fine, D.; Sharma, D.; Torsi, L.; Dodabalapur, A. Anal. Bioanal. Chem. 2006, 384, 310. (c) Torsi, L. Anal. Bioanal. Chem. 2006, 384, 309. (d) Zhu, Z.-T.; Mason, J. T.; Dieckmann, R.; Malliaras, G. G. Appl. Phys. Lett. 2002, 81, 4643. (e) Someya, T.; Katz, H. E.; Gelperin, A.; Lovinger, A. J.; Dodabalapur, A. Appl. Phys. Lett. 2002, 81, 3079. J. AM. CHEM. SOC. 2006, 128, 12851-12869

9

12851

Yoon et al.

ARTICLES

comparable to or surpassing7 those of the common inorganic semiconductor for the aforementioned applications: amorphous hydrogenated silicon (a-Si:H).8 As an example, this material is currently used in fabricating thin film transistors (TFTs) for LC/ LED displays and exhibits an electron carrier mobility of ∼ 1.0 cm2/(V s) and Ion:Ioff ratio >106. For OSCs, such levels of TFT performance have been achieved by continuous innovation in molecular design9 as well as by careful control of vapor-/ solution-phase film growth conditions via optimization of substrate temperature, solvent, deposition/solvent evaporation rate, material purity, etc.10 It is generally accepted that all of the aforementioned factors strongly influence semiconductor film microstructure, which in turn heavily influences charge transport.11 Other critical factors affecting overall semiconductor/ device performance are chemical composition, surface functionalization, and surface morphology (roughness) of the gate dielectric.12,13 Pioneering studies by Horowitz and Garnier14 investigated the consequences for oligothiophene-based TFT response of employing various polymeric insulators, instead of SiO2, as the gate dielectric layer. This work was followed by more recent investigations.15,16 For example, Sirringhaus et al. showed that implementation of appropriate insulators facilitates electron (7) (a) Lee, K. J.; Motala, M. J.; Meitl, M. A.; Childs, W. R.; Menard, E.; Shim, A. K.; Rogers, J. A.; Nuzzo, R. G. AdV. Mat. 2005, 17, 2332. (b) Sele, C. W.; von Werne, T.; Friend, R. H.; Sirringhaus, H. AdV. Mater. 2005, 17, 997. (c) Chabinyc, M. L.; Salleo, A.; Wu, Y.; Liu, P.; Ong, B. S.; Heeney, M.; McCulloch, I. J. Am. Chem. Soc. 2004, 126, 13928. (d) Ling, M. M.; Bao, Z. Chem. Mater. 2004, 16, 4824. (e) Lefenfeld, M.; Blanchet, G.; Rogers, J. A. AdV. Mater. 2003, 15, 1188. (8) (a) Kuo, Y. Thin Film Transistors; Kluwer Academic Publishers: Norwell, MA, USA, 2004; pp 183-202. (b) Sazonov, A.; Striakhilev, D.; Lee, C.H.; Nathan, A. Proc. IEEE 2005, 93, 1420. (c) Hamakawa, Y. Ren. Energy 1996, 8, 10. (9) (a) Videlot-Ackermann, C.; Ackermann, J.; Brisset, H.; Kawamura, K.; Yoshimoto, N.; Raynal, P.; El Kassmi, A.; Fages, F. J. Am. Chem. Soc. 2005, 127, 16343. (b) Wex, B.; Kaafarani, B. R.; Schroeder, R.l; Majewski, L. A.; Burckel, P.; Grell, M.; Neckers, D. J. Mater. Chem. 2006, 16, 1121. (c) Crouch, D. J.; Skabara, P. J.; Lohr, J. E.; McDouall, J. J. W.; Heeney, M.; McCulloch, I.; Sparrowe, D.; Shkunov, M.; Coles, S. J.; Horton, P. N.; Hursthouse, M. B. Chem. Mater. 2006, 18, 6567. (d) Zen, A.; Bilge, A.; Galbrecht, F.; Alle, R.; Meerholz, K.; Grenzer, J.; Neher, D.; Scherf, U.; Farrell, T. J. Am. Chem. Soc. 2006, 128, 3914. (e) Sun, Y.; Ma, Y.; Liu, Y.; Lin, Y.; Wang, Z.; Wang, Y.; Di, C.; Xiao, K.; Chen, X.; Qiu, W.; Zhang, B.; Yu, G.; Hu, W.; Zhu, D. AdV. Funct. Mater. 2006, 16, 426. (f) Takimiya, K; Kunugi, Y.; Konda, Y.; Ebata, H.; Toyoshima, Y.; Otsubo, T J. Am. Chem. Soc. 2006, 128, 3044. (g) Zhang, H.; Wang, Y.; Shao, K.; Liu, Y.; Chen, S.; Qiu, W.; Sun, X.; Qi, T.; Ma, Y.; Yu, G.; Su, Z.; Zhu, D. Chem. Commun. 2006, 755. (10) (a) Locklin, J.; Roberts, M.; Mannsfeld, S.; Bao, Z. J. Macromol. Sci. Polym. ReV. 2006, 46, 79. (b) Jurchescu, O. D.; Baas, J.; Palstra, T. T. M. Appl. Phys. Lett. 2004, 84, 3061. (c) Shtein, M.; Mapel, J.; Benziger, J. B.; Forrest, S. R. Appl. Phys. Lett. 2002, 81, 268. (d) Jentzsch, T.; Juepner, H. J.; Brzezinka, K. W.; Lau, A. Thin Solid Films 1998, 315, 273. (e) Dimitrakopoulos, C. D.; Brown, A. R.; Pomp, A. J. Appl. Phys. 1996, 80, 2501. (f) Laquindanum, J. G.; Katz, H. E.; Lovinger, A. J.; Dodabalapur, A. Chem. Mater. 1996, 8, 2542. (11) (a) Naraso; Nishida, J.-i.; Ando, S.; Yamaguchi, J.; Itaka, K.; Koinuma, H.; Tada, H.; Tokito, S.; Yamashita, Y. J. Am. Chem. Soc. 2005, 127, 10142. (b) DeLongchamp, D. M.; Sambasivan, S.; Fischer, D. A.; Lin, E. K.; Chang, P.; Murphy, A. R.; Frechet, J. M. J.; Subramanian, V. AdV. Mater. 2005, 17, 2340. (c) Abthagir, P. S.; Ha, Y.-G.; You, E.-A.; Jeong, S.-H.; Seo, H.-S.; Choi, J.-H. J. Phys. Chem. B 2005, 109, 23918. (d) Porzio, W.; Destri, S.; Giovanella, U.; Pasini, M.; Motta, T.; Natali, D.; Sampietro, M.; Campione, M. Thin Solid Films 2005, 492, 212. (e) Singh, Th. B.; Gunes, S.; Marjanovic, N.; Sariciftci, N. S.; Menon, R. J. Appl. Phys. 2005, 97, 114508/1. (f) Facchetti, A.; Letizia, J.; Yoon, M.-H.; Mushrush, M.; Katz, H. E.; Marks, T. J. Chem. Mater. 2004, 16, 4715. (g) Meng, H.; Zheng, J.; Lovinger, A. J.; Wang, B.-C.; Van Patten, P. G.; Bao, Z. Chem. Mater. 2003, 15, 1778. (h) Videlot, C.; Ackermann, J.; Blanchard, P.; Raimundo, J.-M.; Frere, P.; Allain, M.; de Bettignies, R.; Levillain, E.; Roncali, J. AdV. Mater. 2003, 15, 306. (i) Videlot, C.; Ackermann, J.; El Kassmi, A.; Raynal, P. Thin Solid Films 2002, 403, 380. (j) Chen, X. L.; Lovinger, A. J.; Bao, Z.; Sapjeta, J. Chem. Mater. 2001, 13, 1341. (k) Horowitz, G.; Hajlaoui, M. E. Synth. Met. 2001, 122, 185. (l) Katz, H. E. J. Mater. Chem. 1997, 7, 369. (12) (a) Facchetti, A.; Yoon, M.-H.; Marks, T. J. AdV. Mater. 2005, 17, 1705. (b) Veres, J.; Ogier, S.; Lloyd, G.; de Leeuw, D. Chem. Mater. 2004, 16, 4543. 12852 J. AM. CHEM. SOC.

9

VOL. 128, NO. 39, 2006

transport for typical organic p-type semiconductors. More recently, it was also reported that certain insulators enable ambipolar transport for pentacene- and rubrene-based OTFTs (organic TFTs).17 Regarding OTFT dielectric-semiconductor interfacial effects, most studies have employed SiO2 as the dielectric and pentacene as the semiconductor.18-28 It is generally accepted that gate dielectric surface roughness is an important parameter affecting OTFT electrical performance,18 and it was shown that rougher gate dielectric surfaces result in smaller pentacene grains and lower OTFT carrier mobilities.19 Note that several groups have explored correlations between pentacene grain size (tuned by varying film deposition rate, substrate temperature) and charge mobility, with most reporting increased mobility with increased pentacene grain size, although many aspects remain controversial.20,21 Very recently, the (13) (a) Jung, T.; Dodabalapur, A.; Wenz, R.; Mohapatra, S. Appl. Phys. Lett. 2005, 87, 182109/1. (b) Backlund, T. G.; Osterbacka, R.; Stubb, H.; Bobacka, J.; Ivaska, A. J. Appl. Phys. 2005, 98, 074504/1. (c) Panzer, M. J.; Frisbie, C. D. J. Am. Chem. Soc. 2005, 127, 6960. (d) Maliakal, A.; Katz, H. E.; Cotts, P.; Subramoney, M.; Mirau, S. P. J. Am. Chem. Soc. 2005, 127, 14655. (e) Schroeder, R.; Majewski, L. A.; Grell, M. AdV. Mater. 2005, 17, 1535. (f) Wang, G.; Moses, D.; Heeger, A. J.; Zhang, H.-M.; Narasimhan, M.; Demaray, R. E. J. Appl. Phys. 2004, 95, 316. (g) Dimitrakopoulos, C. D.; Purushothaman, S.; Kymissis, J.; Callegari, A.; Shaw, J. M. Science 1999, 283, 822. (14) Peng, X.; Horowitz, G.; Fichou, D.; Garnier, F. Appl. Phys. Lett. 1990, 57, 2013. (15) (a) Jang, Y.; Kim, D. H.; Park, Y. D.; Cho, J. H.; Hwang, M.; Cho, K Appl. Phys. Lett. 2006, 88, 072101. (b) Dinelli, F.; Moulin, J.-F.; Loi, M. A.; Da Como, E.; Massi, M.; Murgia, M.; Muccini, M.; Biscarini, F.; Wie, J.; Kingshott, P. J. Phys. Chem. B 2006, 110, 258. (c) Nunes, G., Jr.; Zane, S. G.; Meth, J. S. J. Appl. Phys. 2005, 98, 104503/1. (d) Knipp, D.; Kumar, P.; Voelkel, A. R.; Street, R. A. Synth. Met. 2005, 155, 485. (e) Naber, R. C. G.; Blom, P. W. M.; Gelinck, G. H.; Marsman, A. W.; de Leeuw, D. M. AdV. Mater. 2005, 17, 2692. (f) Stadlober, B.; Zirkl, M.; Beutl, M.; Leising, G.; Bauer-Gogonea, S.; Bauer, S. Appl. Phys. Lett. 2005, 86, 242902/1. (g) Klauk, H.; Halik, M.; Zschieschang, U.; Schmid, G.; Radlik, W.; Weber, W. J. Appl. Phys. 2002, 92, 5259. (16) Chua, L.-L.; Zaumsell, J.; Chang, J.-F.; Ou, E. C.-W.; Ho, P. J.-H.; Sirringhaus, H.; Friend, R. H. Nature 2005, 434, 194. (17) (a) Singh, T. B.; Meghdadi, F.; Gunes, S.; Marjanovic, N.; Horowitz, G.; Lang, P.; Bauer, S.; Sariciftci, N. S. AdV. Mater. 2005, 17, 2315. (b) Takahashi, T.; Takenobu, T.; Takeya, J.; Iwasa, Y. Appl. Phys. Lett. 2006, 88, 033505/1. (18) (a) Knipp, D.; Street, R. A.; Volkel, A.; Ho, A. J. Appl. Phys. 2003, 93, 347. (b) Knipp, D.; Street, R. A.; Krusor, B. S.; Apte, R. B.; Ho, J. Proc. SPIE-Int. Soc. Opt. Eng. 2001, 4466, 8. (19) Steudel, S.; De Vusser, S.; De Jonge, S.; Janssen, D.; Verlaak, S.; Genoe, P.; Heremans, J. Appl. Phys. Lett. 2004, 85, 4400. (20) (a) Ruiz, R.; Choudhary, D.; Nickel, B.; Toccoli, T.; Chang, K.-C.; Mayer, A. C.; Clancy, P.; Blakely, J. M.; Headrick, R. L.; Iannotta, S.; Malliaras, G. G. Chem. Mater. 2004, 16, 4497. (b) Knipp, D.; Murti, D. K.; Krusor, B.; Apte, R.; Jiang, L.; Lu, J. P.; Ong, B. S.; Street, R. A. Mater. Res. Soc. Symp. Proc. 2002, 665, 207. (c) Horowitz, G.; Hajlaoui, M. E. AdV. Mater. 2000, 12, 1046. (d) Klauk, H.; Jackson, T. N. Solid State Technol. 2000, 43, 63. (e) Lin, Y.-Y.; Gundlach, D. J.; Nelson, S. F.; Jackson, T. N. IEEE Trans. Electron DeVices 1997, 44, 1325. (21) (a) Yang, S. Y.; Shin, K.; Park, C. E. AdV. Funct. Mater. 2005, 15, 1806. (b) Gundlach, D. J.; Kuo, C.-C.; Nelson, S. F.; Jackson, T. N. IEEE 57th DeVice Research Conference Digest; Santa Barbara, CA; IEEE Publishing Services: New York, USA 1999; pp 164-165. (22) Fritz, S. E.; Kelly, T. W.; Frisbie, C. D. J. Phys. Chem. B 2005, 109, 10574. (23) (a) Salleo, A.; Chabinyc, M. L.; Yang, M. S.; Street, R. A. Appl. Phys. Lett. 2002, 81, 4383. (b) Lin, Y. Y.; Gundlach, D. J.; Nelson, S. F.; Jackson, T. N. IEEE Electron DeVices Lett. 1997, 18, 606. (24) Yagi, I.; Tsukagoshi, K.; Aoyagi, W. Appl. Phys. Lett. 2005, 86, 1035021. (25) (a) Kobayashi, S.; Nishikawa, T.; Takenobu, T.; Mori, S.; Shimoda, T.; Mitani, T.; Shimotani, H.; Yoshimoto, N.; Ogawa, S.; Iwasa, Y. Nat. Mater. 2004, 3, 317. (b) Pernstich, K. P.; Goldmann, C.; Krellner, C.; Oberhoff, D.; Gundlach, J.; Batlogg, B. Synth. Met. 2004, 146, 325. (26) (a) Pernstich, K. P.; Rashid, A. N.; Haas, S.; Schitter, G.; Oberhoff, D.; Goldmann, C.; Gundlach, D. J.; Batlogg, B. Los Alamos Natl. Lab., Prepr. Arch., Condens. Matter 2004, 1-9, arXiv:cond-mat/0410014. (b) Kelley, T. W.; Boardman, L. D.; Dunbar, T. D.; Muyres, D. V.; Pellerite, M. J.; Smith, T. P. J. Phys. Chem. B 2003, 107, 5877. (27) Kelley, T. W.; Muyres, D. V.; Baude, P. F.; Smith, T. P.; Jones, T. D. MRS Symp. Proc. 2003, 771, 169. (28) (a) Deman, A. L.; Tardy, J. Mater Sci. Eng. C 2006, 26, 421. (b) Itoh, E.; Miyairi, K. Thin Solid Films 2006, 499, 95. (c) Yoshida, M.; Uemura, S.; Kodzasa, T.; Kamata, T.; Matsuzawa, M.; Kawai, T Synth. Met. 2003, 137, 967. (d) Liu, P; Wu, Y; Li, Y; Ong, B S.; Zhu, S J. Am. Chem. Soc. 2006, 128, 4554.

Organic Field-Effect Transistor Performance Correlations

ARTICLES

Figure 1. Chemical structure of the organic semiconductors (OSCs; left) and monolayers/polymers (right) employed in this study. The bottom gate OTFT device configuration is also shown (center).

interplay among SiO2 dielectric root-mean-square (rms) roughness, pentacene morphology, and charge transport was further studied using variable-temperature charge transport measurements.22 In addition to dielectric morphology, the effects of dielectric surface chemical modification on device performance have been explored using self-assembled monolayers (SAMs).23 In general, simple SiO2 hydrocarbon functionalization using octadecyltrichorosilane (OTS) or hexamethyldisilazane (HMDS) enhances the mobilities and lowers the off-currents of most OSCs.24 More recently, it was shown that threshold voltage as well as mobility can be modulated by SAM dipole induced builtin surface potentials.25 Although these results demonstrate the importance of controlling fundamental dielectric and/or interfacial properties for optimizing/controlling OSC charge transport efficiency, connections among dielectric bulk/surface chemistry, surface energy, dielectric-semiconductor interfacial morphology, and electrical properties are not completely clear,26 particularly for far less developed electron-transporting organic semiconductors. An attractive alternative approach for modifying dielectric surfaces is to deposit a second layer, such as a spin-on polymer, on top.27 The major advantages of this approach are that, in contrast to SAM functionalization of oxide insulators, film deposition is not limited by the chemistry required for silane/ phosphate coupling to the dielectric surface and different film thicknesses are readily accessed. Consequently, this methodology allows effective modification of bottom layer topography, hence planarization of rough dielectric surfaces, while simultaneously fine-tuning the chemical properties of the dielectric surface via choice of the deposited polymer. Despite this great

potential, to our knowledge investigations using this approach are sparse, limited to pentacene OTFTs. However, these suggest that pentacene carrier mobility can be tuned substantially, typically by a factor of 2-5×.28 In this contribution, we report a systematic evaluation of the OTFT response characteristics of six organic semiconductors grown on four different SiO2polymer bilayer dielectric structures and compare them to HMDS-functionalized and pristine SiO2 dielectrics. The questions we address are whether this bilayer dielectric strategy is a general approach to enhancing charge carrier mobility, whether there is a correlation between semiconductor-polymer combination and OTFT performance, and the nature of the dielectric surface chemical factors underlying such interfacial effects on charge transport. We demonstrate that OTFT mobility for certain semiconductors can be modulated to a very large extent (several orders of magnitude). To address the above questions we selected for TFT fabrication OSCs with very different core structures, chemical functionalities, and frontier molecular orbital (FMO) energies, including those exhibiting hole, electron, and ambipolar transport on pristine/HMDS-treated SiO2 (Figure 1).29 The bottom layer of the bilayer dielectrics investigated in this study is 300-nmthick SiO2 thermally grown on p+-Si because of the ready accessibility of these substrates and the excellent insulating properties with extremely low leakage currents. Indeed, it has (29) (a) Yoon, M.-H.; DiBenedetto, S. A.; Facchetti, A.; Marks, T. J. J. Am. Chem. Soc. 2005, 127, 1348. (b) Facchetti, A.; Mushrush, M.; Yoon, M.H.; Hutchison, G. R.; Ratner, M. A.; Marks, T. J. J. Am. Chem. Soc. 2004, 126, 13859. (c) Facchetti, A.; Yoon, M.-H.; Stern, C. L.; Hutchison, G. R.; Ratner, M. A.; Marks, T. J. J. Am. Chem. Soc. 2004, 126, 13480. (d) Bao, Z.; Lovinger, A. J.; Brown, J. J. Am. Chem. Soc. 1998, 120, 207. J. AM. CHEM. SOC.

9

VOL. 128, NO. 39, 2006 12853

Yoon et al.

ARTICLES

been reported that, in addition to the dielectric surface roughness18 and dielectric constant,30 the leakage current through the gate insulator has major consequences for OTFT function.31 In this regard, a bilayer approach with SiO2 should enable comparisons of the effects of various polymer surface functionalizations without complications of differing gate leakage currents through the different polymer dielectric layers.32 Furthermore, with this approach, transistor performance parameters on different dielectric modifications can be realistically compared over the same range of gate voltage/electric fields and charge densities accumulated at the semiconductordielectric interface, since the relatively thin polymer film can be employed on top of the SiO2 and therefore maintain the total insulating layer thickness in a similar range. Note that polymeric dielectric materials typically require very different film thicknesses (sometimes very thick) to minimize gate leakage, and that it is not rare to observe mobility variations with the gate voltage/electric field associated with these thickness variations. In the present study, the top polymer layer was chosen/modified to provide a wide range of surface/film properties such as the chemical functionality, hydrophilicity, and polymer dielectric constant. Furthermore, the effect of using polymer layers of different thicknesses was investigated. All of the new bilayer dielectrics were characterized by impedance spectroscopy, quantitative leakage current density measurements, advancing aqueous contact angles, and atomic force microscopy (AFM). The effects of dielectric surface modifications on the OSC microstructure were investigated in detail using a combination of techniques including AFM, scanning electron microscopy (SEM), and wide-angle X-ray diffraction (WAXRD). It will be seen that polymer coating of inorganic insulators is a general strategy for strongly modulating electron transport in TFT devices, whereas hole transport is much less affected. This study thus demonstrates correlations between OTFT carrier mobility, relative semiconductor FMO energies, and the chemical nature of the dielectric surface. Insights into the chemical origin of major charge trapping sites are also provided, and it is shown that judicious choice of polymer coatings can “prime” any dielectric surface for organic semiconductor deposition to enhance OTFT performance. Experimental Section Materials. The semiconductors R,ω-diperfluorohexylcarbonylquaterthiophene (DFHCO-4T),29a R,ω-dihexylcarbonylquaterthiophene (DHCO-4T),29a R,ω-diperfluorohexylquaterthiophene (DFH-4T),29b and R,ω-dihexylquaterthiophene (DH-4T)29c were available in our laboratory from previous syntheses, while pentacene (P5) and hexadecafluorocopperphthalocyanine (CuFPc)29d were purchased from Aldrich and purified by multiple gradient vacuum sublimation before use. Polystyrene (PS, Mw ) 280k) and polyvinylalcohol (PVA, Mowiol 40-88, Mw ) 127k) were purchased from Aldrich and used without further purification. Prime grade silicon wafers (p+-Si) with ∼ 300 nm ((5%) thermally grown oxide (from Montco Semiconductors) were used as device substrates. Film Deposition and Characterization. All p+-Si/SiO2 substrates were cleaned by sonication in absolute ethanol for 3 min and then by (30) (a) Veres, J.; Ogier, S. D.; Leeming S. W.; Cupertino, D. C.; Khaffaf, S. M. AdV. Func. Mater. 2003, 13, 199. (b) Jang, Y.; Kim, D. H.; Park, Y. D.; Cho, J. H.; Hwang, M.; Cho, K. Appl. Phys. Lett. 2005, 87, 152105. (31) de Boer, R. W. I.; Iosad, N. N.; Stassen, A. F.; Klapwijk, T. M.; Morpurgo, A. F. Appl. Phys. Lett. 2005, 86, 0321031/1. (32) Jang, Y.; Kim, D. H.; Park, Y. D.; Cho, J. H.; Hwang, M.; Cho, K. Appl. Phys. Lett. 2006, 88, 072101/1. 12854 J. AM. CHEM. SOC.

9

VOL. 128, NO. 39, 2006

oxygen plasma treatment for 5 min (20 W). For the SiO2 coating layer, -SiMe3 groups were introduced using hexamethyldisilazane (HMDS), deposited by placing the SiO2 substrates in an N2-filled chamber saturated with HMDS vapor for 36-48 h. PS (5.0, 7.5, 15, or 30 mg/ mL in anhydrous toluene), crosslinked-polystyrene33 (CPS, 1:1 volume mixture of PS 7.0 mg/mL and 1,6-bis-trichlorosilylhexane in toluene), and PVA (30 mg/mL in Millipore water) were spin-coated onto substrates at 5000 rpm in the air (relative humidity ∼30%) and cured in a vacuum oven at 80 °C overnight. For PS-Ox coating, PS1 films were exposed to an oxygen plasma for a minimal time (5 s, 20 W) before characterization and subsequent semiconductor deposition. Film thicknesses were measured by profilometry (Tencor, P10). Atomic force microscopic (AFM) images including rms roughness were obtained using a JEOL-5200 Scanning Probe Microscope with silicon cantilevers in the tapping mode, using WinSPM Software. For capacitance measurements, metal-insulator-semiconductor (MIS) structures were fabricated by depositing gold electrodes (200 µm × 200 µm) on the polymer-coated p+-Si/SiO2 substrates. All of the semiconducting materials were vacuum deposited at 2 × 10-6 Torr (∼500 Å thickness, 0.2 Å/s growth rate) while maintaining the substrate temperature at ∼50 °C. Thin films of organic semiconductors were analyzed by standard wide angle θ-2θ X-ray film diffractometry (WAXRD) using monochromated Cu KR radiation. Semiconducting films were coated with 3 nm of sputtered Au before scanning electron microscopic (SEM) imaging using a Hitachi S4500 FE microscope. For FET device fabrication, top-contact electrodes (∼50 nm) were deposited by evaporating gold (3 × 10-6 Torr) through a shadow mask with the channel length (L) and width (W) defined as 100 µm and 5000 µm, respectively. Electrical Measurements. The capacitance of the bilayer dielectrics was measured on MIS structures using a Signaton probe station equipped with a digital capacitance meter (model 3000, GLK Instruments) and an HP4192A Impedance Analyzer. All OTFT measurements were carried out under a vacuum (1 × 10-5 Torr) using a Keithly 6430 subfemtoammeter and a Keithly 2400 source meter, operated by a local Labview program and GPIB communication. Triaxial and/or coaxial shielding was incorporated into the probe station to minimize the noise level. Mobilities (µ) were calculated in the saturation regime using the relationship34: µsat ) (2IDSL)/[WCi(VG - VT)2], where IDS is the sourcedrain saturation current; Ci is the gate dielectric capacitance (per area), VG is the gate voltage, and VT is the threshold voltage. The latter can be estimated as the x intercept of the linear section of the plot of VG vs (IDS)1/2.

Results

This account begins with a discussion of bilayer dielectric fabrication and surface characterization based on advancing aqueous angle measurements and AFM. For electrical characterization, leakage current and capacitance in MIS structures are then measured and discussed. Next, we describe the fabrication and characterization of OTFTs using various semiconductors and bilayer dielectrics. Finally, semiconductor growth mechanism and film morphologies/microstructures as probed by AFM, SEM, and WAXRD are discussed in the context of correlations between dielectric modifications and semiconductor film properties. Bilayer Dielectric Fabrication and Characterization. All of the bilayer dielectric samples were fabricated on p+-Si/SiO2 (300 nm) substrates. The top polymer layer was deposited by spin-coating according to the procedure described in the Experimental Section. The polymers employed in this study are (33) Yoon, M.-H.; Yan, H.; Facchetti, A.; Marks, T. J. J. Am. Chem. Soc. 2005, 127, 10388. (34) Sze, S. M. Physics of Semiconductor DeVices, 2nd ed.; John Wiley & Sons: USA, 1981.

Organic Field-Effect Transistor Performance Correlations

ARTICLES

Table 1. Properties of the Bilayer Dielectrics Investigated in this Study Including Top-Layer Film Thickness (D, nm), Dielectric Surface RMS Roughness (F, nm), Advancing Aqueous Contact Angle (θ, deg), Effective Areal Capacitance (Ci, nF/cm2), Dielectric Constant (ktop), and Effective Dielectric Constant (keff) dielectric

D

F

θ

Ci

ktop

keff

BARE HMDS PS1 PS2 PS3 PS4 PVA CPS PS-Oxy

0 0.4 24 31 71 150 115 13 24

0.1 0.1 0.2 0.2 0.3 0.3 0.3 0.9 0.3

PS ≈ PVA. Importantly, the highest and lowest degrees of texture in air-sensitive n-type semiconductor films on PS-Ox and PS/PVA, respectively, are not directly correlated with the apparent device performance trends (Table 2, µPVA >> µPS-Ox). Considering that for the air-sensitive n-type semiconductors, each type of semiconductor film on the various bilayer structures is prepared in a single batch leading to the same film thickness, and that the maximum θ-2θ scan intensity variation on different bilayer structures is relatively small (within a factor of 2-5), this provides additional evidence that their transistor performance variation can be ascribed to differences in semiconductor-dielectric interface chemical properties rather than to the film morphology/microstructures (vide infra). In contrast, the intensity of the WAXRD scans of the air-stable n-type and p-type semiconductors on PVA is significantly lower by a factor of 50-100 than those on the other dielectrics as indicated in Figure 10, while the overall maximum intensity variation on various bilayer dielectrics follows the trend PS-Ox > Bare g HMDS > PS >> PVA. Especially, in the case of CuFPc on PVA, the first-order diffraction peak exhibits not only substantially less intensity but also a twice larger full width at halfmaximum (fwhm ≈ 0.4° in 2θ) than the same semiconductor films on the other bilayer dielectrics, indicating a poorly ordered film microstructure on PVA. The poor crystallinity of CuFPc, DH-4T, and P5 films on PVA as revealed by WAXRD can be correlated with the relatively poor performance of the corresponding TFT devices (Table 2). In addition to the relatively low degrees of film texturing, DH-4T and P5 films on PVA exhibit different molecular orientations from those observed on the other dielectric surfaces. DH-4T WAXRD scans exhibit a single set of reflections; however, the d-spacing calculated for films grown on PVA (29.3 Å) is significantly larger than that found on all of the other substrates (28.3 Å), demonstrating a different growth mode. WAXRD θ-2θ scans of the P5 films reveal an interesting dielectric-promoted microstructural transition, the details of which will be discussed elsewhere.51 Briefly, P5 films on PS1 and PS-Ox are characterized by an almost phase-pure film, with a d-spacing of 15.4 Å, whereas on PVA evidence of a different single phase is observed with d ) 14.5 Å, and these two d-spacings correspond to the so-called, previously identified “thin film” and “bulk” pentacene phases, respectively.18a Note that the WAXRD scans of P5 films on HMDS and Bare exhibit the presence of both phases with comparable diffraction intensities. To a greater extent, the films of the p-type materials on PVA are characterized by Bragg progressions with far smaller intensities and broader widths than those of the same (51) Kim, C.; Facchetti, A.; Marks, T. J. Manuscript in preparation. 12864 J. AM. CHEM. SOC.

9

VOL. 128, NO. 39, 2006

thickness grown on the other substrates, indicating that both P5 and DH-4T on PVA exhibit less ordered film microstructures. To summarize these observations, the morphology and microstructure within the air-sensitive n-type/ambipolar semiconductors are rather insensitive to the dielectric layer surface, whereas the TFT charge transport is extremely sensitive. In contrast, the air stable n-type and p-type materials exhibit relatively modest dielectric-related TFT performance alterations despite the much greater variations in semiconductor film morphology, crystallinity, and molecular orientation. Discussion

The present results for OTFTs fabricated on bilayer dielectrics having diverse surface chemical functionalities and using a broad range of organic semiconductor types provide new insights into critical relationships between OTFT performance parameters and semiconductor molecular/electronic structure, semiconductor film microstructure evolution, growth mechanism, and dielectric surface characteristics. Although it is not always possible to unambiguously differentiate among the relative contributions of all of these interconnected effects, this investigation provides clear evidence as to which of the aforementioned effects dominates for devices fabricated with most semiconductordielectric combinations. In the following discussion we focus on relationships between OTFT performance and dielectric surface chemistry. Since unfunctionalized SiO2 is the substrate from which all of the functionalized/bilayer dielectrics are fabricated, we use this insulator as a reference point to better understand the origins of performance variations. In the case of carrier mobility, to clearly visualize changes, we introduce the enhancement factor η ) µX/µBare defined as the ratio of the field-effect carrier mobility observed for a certain bilayer structure X (µX) and that measured on the Bare (µBare) substrate. These results are plotted in Figure 11A for all of the semiconductors investigated. It can be clearly seen that η increases on average when moving from the sides to the center of the plot, hence from the less air-insensitive n-type and p-type semiconductors to the more sensitive n-type and ambipolar materials. Therefore, there is a distinct correlation between the dielectric surface chemistry and the empirical sensitivity of the semiconductor majority carrier type to ambient. Figure 11B plots the electrochemically derived frontier molecular orbital energy levels (HOMO and LUMO) for the semiconductors employed in this study. Note that the low-lying LUMO and high-lying HOMO molecules are generally those exhibiting the least sensitivity to the nature of the dielectric surface. Consequently, the utility of employing OSCs having progressive variations in MO energies to probe semiconductor-dielectric interfacial properties finds experimental confirmation. The general trend in OTFT mobilties as a function of the various bilayer dielectrics can be summarized by the following observations: (1) For air-stable n-type semiconductors such as CuFPC as well as cyanated perylene derivatives,52 OTFT performance parameters are relatively insensitive to the dielectric (52) We have synthesized [Jones, B.; Ahrens, M. J.; Yoon, M.-H.; Facchetti, A.; Marks, T. J.; Wasielewky, M. R. Angew. Chem., Int. Ed. 2004, 43, 6363] and investigated a number of alkyl-substituted cyanated perylenes and found very little change in carrier mobility on different dielectrics. Jones, B.; Wasieleswky, M.; Marks, T. J. Manuscript in preparation.

Organic Field-Effect Transistor Performance Correlations

ARTICLES

Figure 11. (A) Histogram showing the mobility enhancement ratio η ) µX/µBare and (B) Electrochemically derived FMO energy levels for the organic semiconductors investigated in this study. The crosshatched bars denote semiconductor-dielectric combinations for which the largest variations in semiconductor film morphology are observed.

surface (0.7 < ηCuFPc < 1.1), with the reasonable exception of PVA (vide infra). (2) The n-type mobilities and current onoff ratios of air-sensitive n-type semiconductors such as DFHCO-4T, DFH-4T, and ambipolar DHCO-4T vary substantially (0 < ηn-type < 100) with the nature of dielectric surface, and the performance enhancement is most pronounced in the n-type mobility of DHCO-4T films on PSn dielectrics. (3) The effects of bilayer dielectrics on the p-type semiconducting properties of DH-4T and P5 are minimal (0.8 < ηp-type < 1.3, with the exception of PVA-P5 where ηP5 ) 0.2) even on PSn and PS-Ox, which induce dramatic mobility increases and decreases, respectively, in the air-sensitive n-type devices. In the present study, considering that each semiconductor is simultaneously grown on the various bilayer dielectric layers in a single batch and that great similarities have been demonstrated in semiconductor film morphologies and microstructures, it is reasonable that the observed performance differences for each semiconductor set can be largely attributed to differences in the details of the dielectric-semiconductor interface chemistry. As shown from variable-temperature mobility studies on n-channel organic transistors, the charge transport in these materials is largely limited by poorly understood charge traps,38 in contrast to the coherent bandlike transport operative in most inorganic semiconductors. These traps are thought to be localized at chemical/physical defects, at semiconductor grain boundaries, and/or at the semiconductor-dielectric interface.53 Trap density changes can be estimated by relative VT shifts with respect to that in reference samples (∆ntrap ) (Ci∆VT)/q),54 (53) Newman, C. R.; Frisbie, C. D.; da Silva Filhoqq, D. A.; Bredas, J.-L.; Ewbank, P. C.; Mann, K. R. Chem. Mater. 2004, 16, 4436. (54) Wang, A.; Kymissis, I, Bulovic, V.; Akinwande, I. A. IEEE Trans. Electron DeVices 2006, 53, 9.

where q is the charge on an electron, or by the difference between VT and Von in the same device (ntrap ) [Ci(VT - Von)]/ q).55 Note that VT is a fitting parameter derived from (IDS)1/2 vs VG plots and can vary substantially, depending on the applied gate bias,56 especially when OTFT devices exhibit gate-bias dependent mobility, gate stress effects, and/or hysteresis. In contrast, the subthreshold swing (S) should be less dependent on the aforementioned artifacts.57 Regarding the electrical properties of dielectric-semiconductor interfaces in TFT devices having different dielectric structures, the maximum density of traps can be estimated from eq 3:58

Nmax trap )

[

]

Ci qS log e -1 q kBT

(3)

where kB is Boltzmann’s constant, T is temperature, e is the base of the natural logarithm, and Ci is the areal capacitance of the dielectric structure. The estimated trap densities for the various semiconductor dielectric pairs are depicted in Figure 12. When bulk trap densities in the semiconductor layer are similar among different dielectrics or are far less than interface trap densities, the trend in trap density estimated from eq 3 can be ascribed mainly to trapped charges at the semiconductordielectric interface and used as the maximum estimated interface trap density.57b In the present study, this is a valid assumption (55) Pernstich, K. P.; Haas, S.; Oberhoff, D.; Goldmann, C.; Gundlach, D. J.; Batlogg, B.; Rashid, A. N.; Schitter, G. J. Appl. Phys. 2004, 96, 6431. (56) Ryu, R.; Kymissis, I.; Bulovic, V.; Sodini, C. G. IEEE Electron DeVice Lett. 2005, 26, 716. (57) (a) Unni, K. N. N.; Dabos-Seignon, S.; Nunzi, J.-M. J. Mater. Sci. 2006, 41, 317. (b) Rolland, A.; Richard, J.; Kleider, J.-P.; Mencaraglia, D. J. Electrochem. Soc. 1993, 140, 3679. (58) McDowell, M.; Hill, I. G.; McDermott, J. E.; Bernasek, S. L.; Schwartz, J. Appl. Phys. Lett. 2006, 88, 073505/1. J. AM. CHEM. SOC.

9

VOL. 128, NO. 39, 2006 12865

Yoon et al.

ARTICLES

Figure 12. Histogram showing the maximum estimated (according to eq 3) interface trap densities (Nmaxtrap) for various semiconductor-dielectric combinations. The dotted plots for CuFPc denote that this semiconductor exhibits substantial Ioff currents indicating unintentional electron doping. Similarly, the crosshatched lines for DH-4T/PS-Ox and P5/PS-Ox are due to stress at the high Ioff currents (extra mobile holes) recorded for these devices due to dielectric electron trapping (see text).

Figure 13. Schematic diagram of functional group electron trapping efficiency on various bilayer dielectric layers.

for the air-sensitive n-type semiconductors and p-type DH-4T in view of the similarities demonstrated in the semiconductor film properties. Therefore, a recognizable correlation might be expected between the mobility enhancement effects and the estimated trap densities, in turn dependent on the dielectric surface modifications. In the following, we discuss, first, OTFT devices based on high-lying LUMO n-type semiconductors which exhibit significant modulation of device performance via dielectric surface modification, followed by analysis of the airstable n-type and p-type devices. Interestingly, it can be seen that similar trap density patterns are observed among the air-sensitive n-type semiconductors although trap density comparisons using eq 3 are formally valid only for transistors having identical semiconductors. The apparent trend is that PS-Ox exhibits the greatest trap densities followed by Bare or PVA, while CPS and PS1 exhibit the lowest densities. In addition, the trend in interface trap density as a function of bilayer dielectric structure exhibits a close correlation with TFT mobility for the n-type semiconductors having high-lying LUMOs. Thus, in the case of DFHCO-4T, DFH-4T, and DHCO-4T, PS1 devices with the highest 12866 J. AM. CHEM. SOC.

9

VOL. 128, NO. 39, 2006

mobilities exhibit the lowest interface trap densities and PSOx devices with the lowest mobilities exhibit the largest interface trap densities. Although the nature of interface traps is doubtless dependent on intricate microstructural details of the interaction between the semiconducting and dielectric layers, the interface trap density can be qualitatively understood from a chemical perspective (Figure 13). It is known that, in the absence of special surface modifications, Bare substrates exhibit an interface trap density of ∼1012 cm-2 which is principally attributed to interfacial chemical functionalities/species such as Si-OH, in conjunction with adsorbed H2O, and adventitious carbon contamination.59,60 Such chemical defects can affect charge transport by deep-trapping/doping and/or by scattering carriers at the semiconductor-dielectric interface, and this effect is reflected in device performance parameters. Such interfacial effects are important in organic semiconductor-based devices since the charge transport process is believed to occur within the very first few semiconducting layers in proximity to the (59) Alexandrova, S.; Sxekeres, A. Phys. Status Solidi A 2001, 187, 499. (60) Olthuis, W.; Bergveld, P. IEEE Trans. Electr. Insul. 1992, 27, 691.

Organic Field-Effect Transistor Performance Correlations

gate dielectric.15 Note that the chemical functionalities depicted in Figure 13 should preferentially trap electrons (rather than holes) leading to mobility suppression in n-type semiconductors, and that for p-type semiconductors the principal outcome is introduction of extra holes at the semiconductor-dielectric interface for charge balance and a positive VT shift.39 The importance of controlling interface traps was demonstrated in the realization of ambipolar pentacene TFTs. Thus, Ahles et al. reported that substantial n-type activity in pentacene TFTs is enabled by compensating electron traps at the semiconductordielectric interface via introduction of an ultranthin Ca layer between the pentacene and SiO2 layers.61 More recently, Chua et al. reported that n-type activity in known p-type semiconductors (e.g., F8T2) is stabilized by passivating surface electrontrapping silanol groups (Si-OH) on the SiO2 dielectric and suggested that their electron affinity is ∼3.9 eV.16 Note that the LUMO energies of air-sensitive n-type semiconductors DFHCO-4T, DFH-4T, and DHCO-4T lie very close to or above the negative of this value, while that of air-stable n-type CuFPc is far lower (Figure 11B). Indeed, the mobility enhancement observed on passivating trap-active SiO2 surfaces using HMDS vs Bare is significantly more pronounced for semiconductors with higher LUMO energies (DFHCO-4T < DFH4T, DHCO-4T), while CuFPc exhibits very similar mobilities on Bare and HMDS (Figure 11A), implying that electrontrapping on surface-modified dielectrics is related to the ambient sensitivity/LUMO energies of n-type semiconductors and to the chemical nature of the dielectric surface. Note that although it was previously suggested that elusive n-type organic semiconductor behavior is governed by interface traps, especially on SiO216,40, the present analysis is, to our knowledge, the first study to explicitly relate dielectric surface chemistry to OTFT performance enhancement via the semiconductor FMO energetics and interface traps. In the following, we address the suppressed/enhanced device performance on other dielectric structures from the standpoint of surface chemistry and interface trap energetics. The very significant interface trap reduction on PSn is supported by several mechanistic arguments: (1) effective passivation of surface chemical defects with a very smooth morphology and minimal nanoscopic pinholes, (2) possible stabilization of charge carriers at the dielectric interface by aromatic cores. Regarding improved device stability and performance, Me3Si-terminated HMDS-treated dielectric surfaces have been shown to generally enhance OTFT performance.24 This strategy has been extended to dielectric surface modification using long-chain organosilanes such as octadecyltrichlorosilane (OTS).25 The origin of the device performance improvement by this organosiloxane interface modification is doubtless the substantial reduction in electron-trapping silanol density on the SiO2 surface without affecting the dielectric surface morphology/roughness.16 However, it has been shown that silanol groups cannot be completely removed by such selfassembled alkyl layers62 and that the relatively thin selfassembled alkyl monolayers (24 nm vs 1013 cm2) and can be ascribed to the chemical functionalities generated by the O2 plasma treatment.54 Previous studies of O2 plasma-treated PS surfaces using ATR-FTIR and NEXAFS found that the density of oxidized-carbon functionalities, especially carbonyl groups, is significant.35 Carbonyl groups should have substantial electron affinities, and based on the electrochemical reduction potentials71 of acetophenone (-1.99 V), benzophenone (-1.72 V), and methylvinylketone (-1.11 V) versus SCE, the estimated LUMO energies of carbonyl functionalities generated by the O2 plasma should lie within -2.8 to -3.7 eV. This range of LUMO energies is very close to those of the air-sensitive n-type semiconductors, and carbonyl groups on the dielectric surface are therefore expected to strongly perturb electron transport at the semiconductor-dielectric interface by trapping electrons. This effect is most pronounced in the case of DHCO-4T on (70) Teyssedre, G.; Laurent, C. IEEE Trans. Dielectr. Electr. Insul. 2005, 12, 5. (71) Fry, A. J. Syntheric Organic Electrochemistry; Harper & Row: New York, 1972. 12868 J. AM. CHEM. SOC.

9

VOL. 128, NO. 39, 2006

PS-Ox where n-type activity completely disappears. Note also that the estimated trap density on PS-Ox is much greater than that on Bare, indicating that the carbonyl groups more efficiently trap electrons than silanol. In the case of p-type semiconductors, the effect of electron trapping on device performance is primarily the positive shift of VT, and its extent is far more pronounced on PS-Ox than on Bare as shown in Figure 5 (vide infra). In addition to mobility modulation, the current-voltage hysteresis evident in the present transfer plots can be tuned by employing different bilayer dielectrics. Thus, while PVA films are known to act as insulators and exhibit hysteresis which presumably arises from charge storage and polarization, PVA efficacy as a polymeric gate electret has also been demonstrated.40 In contrast, the more remarkable hysteretic behavior on PS-Ox substrates (Figure 5) must be related to a very large density of deep interface traps. In the case of n-type semiconductors on PS-Ox, the interface trap states are estimated from the subthreshold swing to be ∼1013 cm-2, much greater than those on the other bilayer structures and cause serious currentvoltage hystereses by binding electrons in these rather deep traps at the semiconductor-dielectric interface. Also, this type of electron trapping is partially indicated by gate-bias dependent semiconductor mobilities in the n-type TFTs on PS-Ox. As argued from the AFM morphology results on the very thin films of these semiconductors, the observed mobility variations are largely governed by dielectric surface chemistry rather than by gross film morphology or a growth mechanism, and such semiconductor-dielectric interactions are clearly revealed by the interface trap densities. From the discussion of surface-chemistry-related interface trap energetics, it is arguable that the relative insensitivity of airstable n-type CuFPc performance to the dielectric surface chemistry can be ascribed to the very low-lying LUMO. In contrast to the air-sensitive n-type semiconductor devices, airstable CuFPc devices do not exhibit direct correlations between the observed semiconductor mobilities and the (overall high) estimated trap density (>10-13 cm-2), although the trap density variation pattern is similar to those of the air-sensitive n-type semiconductors. The reason is likely an overestimated trap density due to charge trapping in the semiconducting layer by chemical impurities. Note that CuFPc exhibits a very low-lying LUMO energy, also indicating that this semiconductor is vulnerable to reversible/irreversible doping by chemical impurities rather than to the aforementioned interface chemical functionalities. Indeed, CuFPc TFTs exhibit relatively high IDS currents at zero gate bias (Figure 4), and the theoretically estimated charge accumulation density at the semiconductordielectric interface at VG ) 50 V (3 × 1012 cm-2, n ) Q/e ) (Ci × VG)/e) is far lower than the estimated trapped charge density in Figure 12. Therefore, the estimated trap density includes traps both in the semiconducting layer and at the semiconductor-dielectric interface. Considering that CuFPc films on different dielectrics exhibit different morphologies, the observed mobility modulation in the various bilayer dielectric structures can be attributed to the combined effects of bulk semiconductor film morphology/doping and the semiconductordielectric interface. The estimated trap densities of the present p-type semiconductors, except on the PS-Ox substrates, are almost constant, which is opposite to the air-sensitive n-type semiconductor

Organic Field-Effect Transistor Performance Correlations

trends (Figure 12). Since DH-4T exhibits essentially the same film morphology and microstructure on all of the present bilayer dielectrics, the similarity in calculated trap densities can be ascribed to invariant semiconductor-dielectric interfacial properties. Therefore, the nearly constant hole mobilities of DH-4T on the different dielectrics can be correlated with the relative insensitivity of hole transport to the dielectric surface nature. Note that in the case of DH-4T on PS-Ox, which exhibits much greater interface trap density than those on other dielectric surfaces, Von and VT are shifted positively compared to that on PS1 and PVA. The off current is also significantly increased due to the extra holes created by electrons filling the interface traps, and the relatively large hysteresis can be explained in a similar way.39 In contrast, the estimated trap densities in pentacene devices do not offer a clear picture of interface trapping. Although similar trends in estimated trap densities are apparent in pentacene devices on the bilayer dielectric structures, differences in film morphologies and microstructures revealed by SEM and XRD indicate that the estimated trap densities are affected by the grain boundaries in the semiconducting layer, molecular orientation, degrees of texture, and the surface chemistry of the semiconductor-dielectric interface. For example, although pentacene TFTs on PVA exhibit relatively low trap densities, which are very similar to those on other bilayer structures except PS-Ox, OTFT performance is poorest even compared to that on PS-Ox which exhibits very high trap densities. As demonstrated by the SEM and advancing aqueous contact angle data, such inferior device performance can be principally attributed to small, discontinuous grains in the semiconducting film. Therefore, unlike the air-sensitive n-type semiconductors and DH-4T, which exhibit relatively similar solid-state film behavior on different bilayer dielectrics, assessment of trap densities here involves changes in semiconductor properties such as grain sizes as well as in the semiconductordielectric interface. For the semiconductor-dielectric combinations described here, we are currently performing admittance (ac) measurements in metal-insulator-(organic)semiconductor capacitors over a large range of applied bias and frequency to achieve a quantitative distribution of the interface trapping states as a function of energy.72

ARTICLES

Conclusions

The present contribution describes a general approach to probe OTFT semiconductor-dielectric interface chemical effects on transistor performance parameters using tailored bilayer dielectrics. Very different organic semiconductors with p-/ntype and ambipolar charge transport characteristics are grown on six different bilayer dielectric structures and systematically characterized by AFM, SEM, advancing aqueous contact angles, and WAXRD. In concert, the corresponding transistor device response parameters are investigated in detail. Polystyrene coatings on SiO2 with minimal gate leakage and surface roughness, significantly enhance the mobilities of air-sensitive n-type semiconductors, while such kinds of device performance improvement is nominal in the case of air-stable n-type and p-type semiconductors. Based on interface trap density estimations on the various bilayer structures, electron trapping at the semiconductor-dielectric interface is identified as the origin of the mobility sensitivity to the different surface chemistries in the n-type semiconductor systems having high-lying LUMOs. The present semiconductors generally exhibit very similar film morphologies and microstructures, regardless of the dielectric surface modification. This result demonstrates that controlling dielectric interface chemistry is critical in achieving good n-type performance in OTFT configurations and that the proper choice of n-type semiconductor can be utilized to probe interface properties such as trap types and densities on various dielectric surfaces, without perturbation of the semiconducting film morphology and microstructures on top. Acknowledgment. We thank AFOSR (STTR FA 9550-040080), the NSF-MRSEC program through the Northwestern Materials Research Center (DMR-0076097) and the NASA Institute for Nanoelectronics and Computing (NCC 2-3163) for support of this research. Supporting Information Available: Figures S1-S6. This material is available free of charge via the Internet at http://pubs.acs.org. JA063290D (72) (a) Torres, I.; Taylor, D. M.; Itoh, E. Appl. Phys. Lett. 2004, 85, 314. (b) Saks, N. S.; Ancona, M. G.; Chen, W. Appl. Phys. Lett. 1992, 60, 2261. (c) Nicollian, E.; Goetzberger, A. Bell Syst. Tech. J. 1967, 46, 1055.

J. AM. CHEM. SOC.

9

VOL. 128, NO. 39, 2006 12869